If it were me, I’d choose tools built for RAG (retrieval‑augmented generation) so the AI actually reads the PDF and answers from it, not its own training. Options like ChatGPT with file upload + custom context, Claude with chunked document input, and Vector‑DB systems (Pinecone, Weaviate) + LLM are what most pros use. These let you ask questions about ebooks and get answers grounded in the actual text rather than generic knowledge.